Notes

With the advent of telematics, pricing and underwriting for certain car insurance companies has evolved into a more complex, data-rich task, with a greater focus on driver behaviour as opposed to other driver characteristics. However, a key element to address in using behaviour as an underwriting tool is considering context in which driving decisions are made. This presentation explores optimal self-driving car algorithms as a possible continuous benchmarking and comparison tool against collected driver data. The talk will be divided as follows:

1. Introduction - A look at the current state of telematics-based modeling (approach, implementation, goals).
2. The Importance Of Context - An argument for why data collected should not be considered and modeled in isolation.
3. Benchmark Algorithms in Devices - Using a hypothetical self-driving car algorithm as a yardstick to judge situational driver decisions.
4. Other Uses of Modeling and Analysis - A look at possible useful by-products of these comparisons (causal analysis).